The properties of the regression of standard length on the length of the exopodit of the uropod of Euphausia superba as a method to estimate the distribution of the standard length from penguin stomach contents samples are explored. The distance between the estimated and observed distributions was measured with the Kolmogorov–Smirnov statistic. Sample variability was determined using resampling techniques. The ability of linear functions (Models I and II), allometric and non-parametric methods to recover the observed distributions was evaluated within samples and exchanging parameters between samples. Linear and allometric models proved inadequate to recover the distribution, while the differences obtained with non-parametric methods fall within the bounds of sample variability, suggesting that a linear equation does not capture the relation between total and exopodit length. The use of a non-parametric regression is recommended to increase the sample size when estimating the distributions of prey lengths in the stomach contents of Antarctic penguins.